|Publication number||US7668773 B1|
|Application number||US 10/037,563|
|Publication date||Feb 23, 2010|
|Filing date||Dec 21, 2001|
|Priority date||Dec 21, 2001|
|Publication number||037563, 10037563, US 7668773 B1, US 7668773B1, US-B1-7668773, US7668773 B1, US7668773B1|
|Inventors||Ronald Earl Pruitt|
|Original Assignee||Placemark Investments, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (23), Non-Patent Citations (4), Referenced by (51), Classifications (5), Legal Events (4)|
|External Links: USPTO, USPTO Assignment, Espacenet|
The present invention relates to asset management systems and, more particularly, to asset management systems enabling distributed management and optimization of financial portfolios for after-tax returns.
The financial investment industry has long recognized the need for tax-advantaged, investment products. However, the investment industry has traditionally managed most taxable and tax-exempt financial portfolios in a similar fashion, even though it is clear that trades that are considered good for tax-exempt portfolios may not be appropriate for taxable portfolios.
In the current investment market environment, investors seek to capitalize on economic growth primarily through mutual funds and individual stocks. The growth in both of these products over the last two decades affirms the widely held belief that it is important to have exposure to the equity markets in any diversified asset allocation program. Mutual funds have traditionally catered to the broad retail market, where investors seek investment advice and active management. However, in mutual funds, the opportunity to manage for tax efficiency is virtually eliminated, as tax-exempt and taxable investors are combined into the same pool of assets. Dilemmas for the taxable investor include 1) the inability to restrict short-term capital gains; 2) the hazard of buying into yet unrealized gains; 3) the lack of control over the timing of gain realizations; and 4) the inability to pass losses from a fund to offset taxable events in the rest of the investor's portfolio.
Indeed, a significant downside to investing in mutual funds is the tax treatment of gains and losses. Market volatility and the movement of assets in and out of funds have resulted in many funds generating significant capital gains distributions. When individuals own mutual funds, they receive distributions for all the gains realized through trading. This means that even if the investor does not sell any fund shares, he or she may have to pay taxes on the gains that the fund manager has realized. Indeed, most alarming is the situation that may occur when an individual investor invests after a fund has accumulated significant unrealized gains. If the fund then has a price decline, causing other investors to exit the fund, the fund manager must sell stock to generate the cash needed to pay out selling shareholders. The result is that the new investor can actually lose money, and owe taxes on gains that he or she never received. These tax issues make it advantageous for individual investors to own individual securities, although most will continue to seek the professional management, reduced (or shared) costs, and diversification that is inherent in mutual funds.
By owning individual equities, investors do not inherit other investors' capital gains as they do in mutual funds. For investors seeking a managed account solution, so-called wrap accounts provide a managed financial portfolio to individual investors. However, financial services offering management of individual portfolios have only been available to high net worth investors. Even as to most individually-managed, separate account programs, however, managers typically consider tax implications at a very high level, ignoring each clients cost bases associated with their holdings, changes in tax rates and other relevant tax information specific to each client. Constructing a financial portfolio involves considerable time and effort. For this reason, most managers of wrap programs manage all client portfolios for a given style in the same manner, by constructing a model portfolio that is adhered to for all accounts. While it may be possible to apply certain rules in constructing individual client portfolios (for example, no sin stocks), accounts are generally not customized for each client based on individual tax situations. Tax considerations, where taken into account at all, are typically made across all client accounts at a model portfolio level. Moreover, even the large majority of managers and wrap programs that claim to be tax aware or tax efficient generally just minimize turnover, or match gains and losses as they make trades within the model portfolio.
Such a management process ignores the genuine differences among clients, such as account costs positions, tax rates, and other taxable events in their lives (for example, gains and losses outside the specific account being managed). For taxable accounts, these wrap programs do not provide tax-optimal investment management for individual investors. To optimize after-tax returns for each client, an investment program must consider many circumstances specific to the client—e.g., the investor's tax bracket, gains or losses outside the managed funds to match against investments in the managed portfolio, the respective cost bases for equities in the portfolio, etc. Integration of all necessary information into portfolio management has been a labor-intensive process. Accordingly, due to the complexity involved in maximizing after-tax returns, only multi-million dollar financial portfolios have received the individualized attention required to make the best trades for each individual taxable client.
Market pressure from individual investors, the growth of Internet-delivered solutions, and reduced transaction costs for trading, accounting and other support functions are driving the development of a new category of low-cost, individually diversified portfolios. The emergence of these products is proving to be a viable alternative for individual investors who want to avoid the tax issues of mutual funds and manage their own accounts. The products help individual investors select pre-packaged “folios” or “baskets” of stocks to meet their investment objective. While these products allow individuals to achieve diversification benefits that are normally only available in mutual funds, they are generally either passive products, or must be managed by the individual investor. For those that seek management from established money managers, these new basket products do not provide a solution.
In light of the foregoing, a need exists for asset management systems that enable tax-optimized management of individual financial portfolios in an efficient and cost effective manner. For instance, a need exists for asset management systems that allow for tax-optimized managed account products that are available to lower net worth investors.
The present invention provides methods, apparatuses and systems allowing for optimization of individual client financial portfolios balancing risk, return, as well as transaction and tax costs, based on the inputs of at least one investment style manager. The present invention enables, in one embodiment, allocation of the management of a financial portfolio among a plurality of managers and the simultaneous optimization of the financial portfolio across all managers for after-tax returns. One embodiment of the present invention allows a professional money manager to deliver investment advice to potentially thousands of individual accounts, but tailors actual buy/sell recommendations to each individual investor's tax position and preferences.
In one embodiment, the present invention integrates inputs from financial advisors, individual investors, money managers, and an accounting system that tracks the cost basis and account holdings in all accounts for each individual investor and determines the optimal decisions for each investor's portfolio. In one embodiment, the present invention allows for the consideration of the investor's tax status and preferences, as well as the tradeoffs between the inputs provided by the money managers and the tax implications of all proposed portfolio adjustments.
In one embodiment, the present invention combines information from professional money managers with client settings gathered by an advisor, and tax-Level portfolio data from an account management system, to derive a customized, managed portfolio of equities optimized for after-tax returns. A particular money manager can make changes to equity selections or other model style settings as appropriate, and, as the portfolios for all clients associated with that money manager are re-optimized, the most recent stock selections and portfolio construction rules are applied. Likewise, an advisor may input changes to a client's tax position as frequently as necessary. Information regarding the client's total tax situation, not only within the managed portfolio, but across the client's assets (for example, gains and losses in other investments), and changes in income tax and capital gains tax rates can all be taken into account when developing the optimal portfolio for each client. The decisions as to which equities to buy or sell in an individual investor's account vary, based on each individual's circumstances. For example, an investor with a higher cost basis might sell a security, based on a moderate “sell” signal from the investment style manager, whereas another investor with a very low cost basis on that same security might hold it because of negative tax consequences that are specific to that investor. The portfolio optimization platform, in one form, optimizes how the money managers' inputs are applied to each individual investor's account, reviewing each account and making individual investment decisions, account by account. By periodically reapplying the advice of respected money managers with current information from the individual investor, the portfolio optimization platform achieves its objective of constructing, and continually reconstructing, the best portfolio of securities for each individual's specific tax position.
As discussed above, the present invention integrates inputs from investment clients, advisors, investment style managers and accounting systems to manage tax-optimized financial portfolios. The present invention may operate in connection with various types of financial portfolios, including managed accounts, wrap accounts, folios, and traditional brokerage accounts. Each investment style manager defines and maintains at least one investment style, essentially managing a model portfolio (e.g., stock selection, portfolio construction, and ongoing management), and provides investment style data including, for example, portfolio construction parameters such as stock selection and ranking data and, in one embodiment, guidance as to a set of appropriate settings defining how the model portfolio should be constructed. However, investment style managers can provide a wide variety of inputs.
According to one embodiment of the present invention, an advisor works with an investment client to determine an appropriate asset allocation for his or her financial portfolio. Within each asset class, the advisor also assists the investment client in selecting at least one appropriate investment style and an initial allocation of funds to each account associated with the selected investment styles. The accounts corresponding to the selected investment styles make up the client investor's portfolio. The account management system maintains each investment client's financial portfolio (in one embodiment, as a collection of separate accounts) and provides necessary inputs to the portfolio management platform. As to each investment client, the portfolio optimization platform combines data relating to the investment client's individual financial circumstances and account settings with the portfolio construction parameters submitted by selected investment style managers to generate tax-optimized, suggested adjustments to the financial portfolio of the investment client. In one form, the resulting suggested adjustments are ultimately transmitted to a middleware application for execution.
A. Advisors and the Advisor Platform
Advisors assist client investors in determining the appropriate investment policy, selecting appropriate investment vehicles, allocating capital to the vehicles, as well as facilitating the sales process via proposal generation and sales management tools. Advisors also provide ongoing management on behalf of the client and deliver performance reporting. Advisors, in one embodiment, essentially acts as distributors of the financial product(s) enabled by the present invention—e.g., tax-optimized, managed investment portfolios.
In one embodiment, an advisor (for example, a Certified Financial Planner) works with an individual client to allocate the assets a client wishes to have managed among a group of professional money managers (here, investment style managers), based upon the investor's specific circumstances. The advisor asks the client standard “risk profiling” questions regarding investment goals, risk tolerance, existing core stock holdings, and tax status. The advisor then creates an asset allocation and offers the investment client a combination of investment styles, each investment style mapping to a specific investment style manager, to achieve the suggested asset allocation. In another embodiment, part or all of the advisor's role is performed by software modules residing on a server connected to the computer network. The application includes functionality allowing for registration and profiling of client investors over the Internet 90, creation of asset allocations, selection of investment styles or managers within each asset class, and the receipt and storage of other data relating to client investors.
Advisor platform 20, in one embodiment, includes tools to support an advisor's role in the system. As
Ongoing management module 27 facilitates the input, storage, retrieval and modification of data relating to each client investor's portfolio and external financial circumstances. In one embodiment, investor database 164 stores account data for each client investor, including an investor identification, address information, password or other authentication data, risk profiling data, asset allocation data, selected investment styles/financial products, capital gains tax rates, tax-based restrictions, restricted securities, current cash needs and timing, portfolio re-balancing preferences, and external financial event data, such as assets held, or gains/losses incurred, outside of the funds managed by the system. As
1) external gains and losses: information relating to gains and losses resulting from financial events having tax implementations outside the managed funds;
2) un-managed assets: financial assets outside/external to the managed portfolio;
3) capital gains tax table: tax rates specific to the client investor that would apply to investment income and capital gains (long/short term); in one embodiment, the tax table information is automatically generated based on other financial data gathered from the client investor and generally available tax tables;
4) tax based restrictions: client based/specified preferences as to what types of taxable events they prefer to avoid (e.g., no investment income, no short term gains, etc.);
5) restricted securities: list of off-limits securities (e.g., no “sin” stocks); in one embodiment, restricted securities data also include any security external to the managed portfolio that the client investor sold for a loss (see below);
6) cash needs & timing: information specific to each client investor as to his or her plans for withdrawals from the managed portfolio; and,
7) re-balancing preferences: settings as to how far the portfolio is allowed to deviate from the initial asset allocation or the duration of time since the last rebalancing, before re-balancing of the portfolio is effected.
In one form, the client investor's information is updated by the advisor and/or the client investor as frequently as necessary.
Client access module 29 allows investor clients remote access to portfolio data and allows investor clients to change such data. In one embodiment, client access module 29 includes web server functionality allowing investor clients access to portfolio account and reporting data over Internet 90. Through client access module 29, client investors can view the holdings and status of their respective portfolios, as well as change portfolio data, such as mailing address and the like, and transfer funds into and out of their portfolios.
In one embodiment, a password is required to gain access to investor portfolio accounts. According to one embodiment, each client portfolio account includes a user name, and a password or an encrypted representation thereof. In one embodiment, client access module 29 authenticates users in a conventional manner. In one embodiment, client access module prompts the user for a user name and a password to authenticate the user. Numerous authentication protocols are known in the art. The actual authentication protocol used is not critical to the invention. In one form, communication of account data to the client investor 30 is transmitted using a secure communications protocol. In one embodiment, the browser implemented on the client investor's computer supports the SSL (“Secure Sockets Layer”) protocol, the S-HTTP (“Secure HTTP”) protocol, or any other similar protocol for transmitting confidential or private information over an open computer network 90.
B. Account Management System
Account management system 80 maintains client investor portfolios and assets corresponding to the client investor portfolios. In one form, a financial institution, such as a brokerage firm, maintains account management system 80, which includes functionality allowing for tax lot accounting, and trade execution. In one form, account management system 80 comprises a middleware system including functionality directed to tax lot accounting, clearing and custodial integration, and trade execution. Account management system 80 can be any accounting system that facilitates trading of accounts, allocation of trades made at a block level to each individual account, ongoing reconciliation of the accounting system with the custodian's records, specific identification of shares sold by tax lot and ongoing tax lot level detail of the cost basis, purchase dates, individual tax lot status, and performance reporting consistent with industry standards. Known systems that meet some or all of the above requirements include Security APL, Portia, dbCAMMS, Centerpiece, Advent (Axys and Moxy) and Eagle Investment Systems (STAR and PACE). As
In one embodiment, as discussed above, a particular client's financial portfolio comprises a plurality of separate accounts relative to account management system 80, where each account maps to a specific investment style manager chosen by the client. Any mutual funds, money market and other cash funds, however, are combined into one account. For certain types of account management systems, this embodiment facilitates auditing of the financial portfolio per investment style manager and the monitoring and reporting of the performance of each investment style manager. However, as discussed below, this configuration requires that allocation parameters be adjusted accordingly (see below). In other embodiments, each client's entire financial portfolio is represented and handled as a single account.
C. Investment Style Managers and Investment Style Management Platform
An investment style manager defines and maintains one or more investment styles, such as large cap growth, small cap value, international, and so on. Each investment style includes all of the portfolio construction and management considerations that a money manager would normally develop to manage a mutual fund or similar investment portfolio. The investment style manager, in one embodiment, initially defines a model portfolio and subsequently adjusts it as necessary. The investment style manager further defines and adjusts such investment style parameters as equity screening requirements (e.g., minimum market cap, price to earnings ranges, sector weightings, turnover constraints, maximum/minimum numbers of equities, etc.), risk modeling parameters, and individual equity recommendations or alpha weightings. The investment style manager submits such investment style data (model portfolio and parameters) to portfolio optimization platform 40 to enable calculation of adjustments to individual portfolios some of whose assets have been allocated to that investment style. In one embodiment, investment style managers submit investment style data on a periodic basis (e.g., daily, weekly, monthly) or in response to changing market or other conditions.
Investment style managers may generate investment style data using their own internal investment processes and provide them via any suitable data transfer technology, such as e-mail, FTP, etc. Moreover, the investment style data may be transferred using any suitable secure transmission technology, such as SSL, HTTP-S, or Virtual Private Networking (VPN) technologies. However, investment style management platform 60, in one embodiment, includes functionality facilitating each investment style manager's role in the system. However, investment style managers may employ any functionality or techniques appropriate to generate the required inputs. As
C.1. Receiving and Processing Investment Style Manager Inputs
In one embodiment, a critical set of inputs into portfolio optimization platform 40 is the security selections of the investment style managers. In a preferred embodiment, investment style managers are required to provide and regularly update their security selections. In one embodiment, investment style managers provide the security selections in the form of a model portfolio as would be appropriate for a tax-exempt client, taking into consideration, however, that the model portfolio must be implementable in account having the minimum amount of assets allowed by the operators of the system (e.g., $100,000). Therefore, if the minimum account size is $100,000, for example, it would be impracticable for investment style managers to offer model portfolios including 500 securities. Moreover, investment style managers should also consider the yield of securities as being taxable. As more fully described below, portfolio optimization platform 40 then manages the portfolios of client investors to minimize tracking error from the model portfolios associated with each respective portfolio and balances that tracking error with the associated transaction and tax costs of trading.
C.1.a. Types of Input and Levels of Participation
In one embodiment, portfolio optimization platform 40 concurrently supports multiple levels of participation by investment style managers. The levels are differentiated by the degree of detail of investment style data provided by the investment style manager. The types and format of the data submitted by investment style managers can vary according to the level of participation desired by each investment style manager. In addition, the investment style data can be communicated as Microsoft Excel files, .CSV files, delimited text files, or any other suitable data format. The delivery can be via secured e-mail, secured file transfer protocol (ftp), or any other suitable data transfer method and/or format, such as XML and SOAP technologies. In one embodiment, investment style managers transmit investment style data to portfolio optimization platform 40 via extranet 93. In one embodiment, portfolio optimization platform 40 includes web server functionality presenting custom-tailored page-based interfaces that facilitate the uploading of investment style data.
1) Model Portfolio
In one embodiment, the minimum participation requirement is the provision of a model portfolio including an identification of each security (e.g., ticker symbol) and a corresponding position weight (see Table 1).
File #1: Model Portfolio
. . .
. . .
. . .
2) Opinion on Legacy Assets
According to a higher level of participation, the investment style manager provides a model portfolio plus an “opinion” on manager-driven “legacy assets” (see Table 2). The investment style manager's opinion, in one embodiment, takes the form of a ranking. A ranking is any qualified measurement of the preference for holding a security. This ranking can take the form of any ordinal data set (numerical ranking such as 1-10 and/or categorical ranking system such as strong buy to strong sell), a continuous data set such as an alpha ranking or a series of price targets with a general or specific time horizon. These rankings are used to infer the excess return potential.
This level of input allows for enhanced portfolio optimization. Legacy assets include any asset that a client holds either due to funding an account with securities (“client-driven” legacy asset), not trading out of a security when the model portfolio changes or deviating from the model portfolio (see alternate securities) and then not trading the security either due to restrictions and/or tax purposes. Because new client investors continuously enter the program, there are many different cost bases for the same security as well as different purchase dates. Therefore, when an investment style manager chooses to remove a security from his model portfolio, it may make sense for a particular client portfolio from a tax perspective to hold on to that security to reduce or defer taxes. The investment style manager's opinion data on such legacy assets facilitates this level of optimization for the client portfolio. Ideally, investment style managers maintain research coverage on securities that have recently traded out of their respective model portfolios as it may be beneficial for some clients to hold these securities, for example, until they reach long term capital gains status. The investment managers' respective opinions on such legacy assets allows for an explicit calculation of the tradeoffs between holding the legacy asset versus selling the legacy asset and adhering more closely to the model portfolio.
File #2: Opinion on Securities in Model plus Legacy
3) Alternate Securities
In another, still higher level of participation, investment style managers also provide a set of alternate securities for those in the model portfolio. Alternate securities are securities that could serve as replacements for securities that reside in the manager's model portfolio that have been screened out based on restrictions and preferences made by the client and/or for purposes of avoiding a wash sale (see below). Investment style managers can link the alternate securities to specific model portfolio securities or can provide a list of alternate securities applicable to the entire model portfolio. Alternate securities can be further classified by the investment style manager as general alternates or “active risk equivalents.”
Active risk is the deviation from a specified target portfolio. Most simply explained, if there is a model portfolio (e.g., an index or a manager model portfolio) and the corresponding security weights are exactly replicated in a clients portfolio then there is no active risk that the client's realized performance will deviate from that of the target. In reality, transaction costs and the timing between when the model portfolio changes and when the client portfolio changes will, at a minimum, result in slight deviations. Whenever there is a deviation from the model, there is active risk; tracking error is the sum of the active risk from each security.
In addition, an investment style manager typically manages a model portfolio in relation to an appropriate benchmark (e.g., S&P 500, Russell 2000 Growth, etc.) wherein the model portfolio's deviation from the benchmark results in tracking error. An active risk equivalent is an alternate security that, when used in place of a particular model portfolio stock, the model portfolio provides a substantially equivalent degree of tracking error from the benchmark as compared to the original model portfolio. Furthermore, a general alternate security is a security that the investment style manager feels is a favorable selection for a client portfolio, but does not meet the more stringent requirements of an active risk equivalent.
As described more fully below, the addition of a set of alternate securities allows greater ability to deliver customized, tax-optimized portfolios to client investors that reflect the best ideas of the investment style managers. Some examples of the need for alternate securities are: 1) deviation from the model to facilitate gain/loss matching across investment style managers; 2) deviation from the model portfolio to accommodate security restrictions imposed by the client; 3) deviation from the model portfolio to facilitate the avoidance of wash sale transactions within and across investment style manager model portfolios.
4) Opinions on Securities in Research Universe
In yet another higher level of participation, investment style managers provide opinion data as to securities in their respective research universes (see Table 3). A research universe is a broad set of securities that make up the asset class a particular investment style manager is focused on managing. In most cases, the research universe encompasses a specified benchmark index (such as the Russell 1000 Growth) or based on a screen of securities with certain characteristics (such as all securities with at least 6 quarters of operating history and a market cap in excess of $5 billion). In one embodiment, investment style managers provide rankings for all securities in their respective research universes. This ranking is used to make an explicit tradeoff decision between taxes, return potential, risk and transaction costs associated with trading. Such opinion data allows for higher degrees of customization and tax-optimization for portfolios of taxable investors, because all tradeoffs are explicitly quantified. For example, if a client investor initially funds the portfolio account with stock (a client-driven legacy asset having an associated cost basis) and that stock is not covered by an investment style manager, the portfolio optimization platform must make implicit assumptions about the relative value of that stock over what the investment style manager currently recommends. Additionally, wash sale restrictions may result in choosing an alternate security. An investment style manager that provides rankings on all stocks in the research universe (to support all forms of legacy assets—manager and client driven) and provides alternate securities is able to support explicit analysis of all tradeoffs (e.g., between tracking error from the manager's model portfolio to a benchmark, the differential return potential between all securities and the tax costs to trade back to the model portfolio).
File #3: Opinion on all securities in universe
or price target)
. . .
. . .
. . .
C.1.b. Frequency of Model Portfolio Updates
Investment style managers, in one embodiment, submit updates to their respective model portfolios and/or other investment style data on a periodic (e.g., daily, weekly, monthly, etc.) or on an as-needed basis. The frequency of updates depends on the normal investment process of the investment style manager. To assist investment style managers providing at least legacy asset data, portfolio optimization platform 40 provides a list of all current legacy assets and the number of client portfolios holding each legacy asset and, optionally, the total dollar weight of each legacy asset. This enables investment style managers to prioritize legacy asset coverage to those securities that are more broadly held within client accounts.
C.1.c. Processing of Investment Style Manager Inputs
Given the different levels of participation concurrently available, portfolio optimization platform 40 must process the investment style data to provide a uniform set of inputs to optimization system 110. For example and in one embodiment, portfolio optimization platform 40 selects alternate securities for model portfolios of the investment style managers who do not provide them.
In one embodiment, investment style data processor 154, for each model portfolio security, determines the most similar securities for use as alternates, using one of two approaches, depending on whether the investment style manager provides rankings of securities within his or her investment universe (step 802) (see above). Investment style data processor 154 then determines whether the investment style manager provided a list of alternate securities (step 804). If so, investment style data processor 154 uses the list of alternates provided (step 805); otherwise, investment style data processor 154 selects a list of potential alternates as follows (step 806). If the investment style manager provides rankings of stocks within the investment universe then, investment style data processor 154 selects 50 securities (outside the model portfolio) having alpha values less than, and 50 securities (outside the model portfolio) having alpha values greater than, the alpha value associated with the current security, and within the same economic sector as the model portfolio security. Investment style data processor 154 does not include securities that have an alpha value less than zero; however, if the investment style manager's alpha associated with the model portfolio security is less than zero then alternate securities having alphas less than zero are permitted.
For either the generated or provided list of alternate securities, investment style data processor 154 arranges the list of securities into a 101×1 matrix with the model portfolio security at the top, followed by the 100 potential alternates. Note that the matrix may be smaller depending on the number of potential alternates, if any, provided by the investment style manager. Investment style data processor 154 then retrieves the benchmark factor exposures (J×1 matrix, where J represents the number of factors) associated with the benchmark designated by the investment style manager (step 808). Investment style data processor 154 then selects the factor exposures for each model portfolio and potential alternate security, where the model portfolio securities appear before the potential alternates in a J×101 matrix (step 810). For each security and each factor exposure associated with each security, Investment style data processor 154 subtracts the benchmark factor exposure from the corresponding factor exposure for the security to calculate, as to each factor exposure, how much any given security differs from the underlying benchmark (step 812).
For each security, investment style data processor 154 then collapses the inter-relationships of the factor exposures into a single term to allow for assessment of how much each alternate security differs from the model portfolio security in relation to systematic risk (step 814). In one form, this is accomplished by computing XTΩX, where X represents each security and the column of associated factor exposures (J×1 matrix), XT is the transpose of this matrix (1×J), and Ω is a J×J variance-covariance matrix of factors themselves.
Investment style data processor 154 then computes the how much the Annual Residual Variance for each security should be considered (step 816). In one embodiment, for each security, investment style data processor 154 selects the benchmark weight (i.e., the weight of the security in the benchmark), setting any alternate security weight not in the benchmark to zero. Then, for each security, investment style data processor 154 then computes the value of [(1−benchmark security weight)2×Annual Residual Variance].
To compute a risk value for each security (step 818), investment style data processor 154 calculates the square root of the sum of 1) the value computed in step 816, and 2) the systematic risk (see step 814) associated with each security.
Lastly, to allow for an assessment of the best alternate security, investment style data processor 154 computes a risk-return information ratio comprising the alpha associated with the security over the risk value computed above (step 820). Investment style data processor 154 then subtracts the information ratio associated with the model portfolio security from the respective information ratios of the alternate securities (step 822). Investment style data processor 154 then arranges the securities in an ordered list starting with the alternate securities having the largest value resulting from step 822 (that is, the alternate securities having the highest risk-return information ratio relative to the model portfolio security) and running to the lowest value (step 824).
If the manager does not provide rankings for the investment universe then, investment style data processor 154 selects an alternate security based on the differences from the model portfolio security so as to select the most similar security as described by a single or multi-factor risk model. In one embodiment, selects 100 securities (not in the model portfolio) having a beta value less than, and 100 securities having a beta value greater than, the beta of the model portfolio security, and within the same Compustat sector (see step 832). However, if a manager provides a general alternate list, this is used instead (see steps 830 and 834).
Investment style data processor 154 then retrieves the factor exposures for the model portfolio security (J×1 matrix, where J is the number of factor exposures) (step 836) and constructs a J×200 matrix comprising the factor exposures for each potential alternate security (step 838). Investment style data processor 154 then subtracts the factor exposures associated with the model portfolio security from the corresponding factor exposures associated with each potential alternate security (step 840).
As discussed above, investment style data processor 154, for each potential alternate security, collapses the inter-relationships of the factor exposures into a single term to allow for assessment of how much each alternate security differs from the model portfolio security in relation to systematic risk (step 842). In one form, this is accomplished by computing XTΩX, where X represents each security and the column of associated factor exposures (J×1 matrix), XT is the transpose of this matrix (1×J), and Ω is a J×J variance-covariance matrix of factor exposures themselves.
Investment style data processor 154 then compute a risk value for each potential alternate security by computing the square root of the sum of 1) the Annual Residual Variance for each security and, 2) the corresponding systematic risk value computed above for each security (step 844). The securities having the smallest risk values computed in step 844 are considered the most similar (step 846).
From the list of most similar securities (generated by either of the two methods provided above), investment style data processor 154, in one embodiment, selects at least the three, but as many as six, of the most similar securities (step 850) and, based on the investment style manager setting for active risk equivalents (see step 852), sets the alternate_security_type for all securities (step 854). In one embodiment, investment style data processor 154 ensures that there are at least two alternate securities from a different industry, and at least two securities from a different economic sub-sector.
D. Portfolio Optimization Platform
Portfolio optimization platform 40 collects requisite inputs from advisors, investment style managers, clients, and associated accounting systems and calculates adjustments to client portfolios optimized for after-tax returns. As discussed in more detail below, portfolio optimization platform 40 determines the optimal decisions for each client portfolio based on the tradeoffs among tax status and preferences of the client, the inputs provided by the investment style manager(s) associated with the portfolio, and the tax implications of all purchases and sales.
Scheduler 130 schedules the optimization of client financial portfolios. In one embodiment, calculation of adjustments to a client portfolio occur on a periodic basis based on a setting stored in client account database 164 for each client investor, but may also be scheduled on an as-needed basis in response to changing market conditions, an explicit request from a client, and the like.
Optimization system 110 is the core component of portfolio optimization platform 40 as it calculates adjustments to individual financial portfolios. Process handler 120 manages the overall optimization process and hand tasks off to the various software modules and associated servers in a distributed computing environment. In one embodiment, process handler 120 passes each discrete task to the appropriate module as a message. In one embodiment, such messages are passed using the SMTP protocol; of course, any suitable protocol for transferring files and other data in a distributed computing environment can be used, such as XML and SOAP technologies. File/array builder 122 and data pre-processor 124 operate to retrieve required inputs from the databases described above and build the files and arrays used to construct the optimization problem. Both file/array builder 122 and data pre-processor 124 include several modules executing the functionality more fully described below. In one embodiment, the files and arrays constructed by builder 122 and pre-processor 124 can be stored in a shared file space for access by later processes. Problem builder 126 operates to construct the optimization problem from the files and arrays constructed by file/array builder 122 and data pre-processor 124. Problem solver 128 solves the optimization problem and returns a solution. Problem solver 128, in one embodiment, includes standard optimization software designed to maximize a given objective function in light of a pre-defined set of constraints. Suitable optimization software packages include, but are not limited to XPRESS-MP, offered by Dash Optimization of Bilsworth Northants, UK, FORT-MP offered by OptiRisk Systems of Uxbridge, Middlesex, UK, and Premium Solver Plus 3.5 offered by Frontline Systems of Incline Village, Nev.
The databases described above can be any form of database known in the art (for example, a relational database or flat-file database). In one embodiment, each database has associated therewith a collection of computer programs enabling the storage, modification, and extraction of information in the database. The databases may be stored on any suitable device ranging from personal computers (for small systems) to mainframes (for large systems). In addition, the functionality of the various modules described above is executed by a plurality of servers implemented in hardware or software, or a combination of both. In one embodiment, each server is a programmable computer executing computer programs, comprising at least one processor, a data storage system, at least one input device, and at least one output device.
In pre-processing step 212, file/array builder 122 and data pre-processor 124 retrieve required inputs, such as the model portfolios associated with the current client portfolio, client preferences, current portfolio data, and the like. Data pre-processor 124 screens and processes the inputs in order to facilitate construction of the optimization problem—e.g., what tax rates to use, which transactions should not be conducted for tax code reasons, etc. (see below). As more fully described below, file/array builder 122 constructs the arrays and files used in the optimization. The output of this process is the input for the optimization (step 214) (in one embodiment, a series of arrays or files that get passed via process handler 120 (see
In one form, the review process involves both automated and non-automated aspects. In one embodiment, an automated tax cost module calculates the tax cost of the recommended trades, while an automated tracking error module estimates the tracking error to the model portfolios associated with the account. In addition, a portfolio manager may assess whether the optimization should be re-run in light of intervening circumstances, if any. For example, the price of some securities may have moved from the point in time the optimization was run. Whether the price movement is substantial enough (i.e., whether, based on the type of inputs and the client account, it makes sense to rerun the optimization) is committed to the discretion of a portfolio manager.
Once the review is conducted and the trades are approved, back office integration server 50 creates trade files that can be loaded into the appropriate account management system 80 for execution (step 218). Where possible, the trades are reviewed inside account management system 80 for restrictions, intra-account wash sales, and any other limitations imposed by the advisor, client or other entity as a quality control check and then the trades are committed to a trade desk 86 (step 220).
In one embodiment, the variables and calculations associated with the optimization of a client's portfolio relate to the client's portfolio as a whole. In embodiments discussed above, where each investment style corresponds to a separate account relative to account management system 80, the recommended adjustments to the client's financial portfolio must be converted to a per-account basis. For example, assume for didactic purposes that the current allocation of the client portfolio for a particular investment style manager is 26% of the total portfolio and that the optimization yields a variable that indicates holding 1.35% in IBM for manager 1 (the percentage based on the total portfolio). This percentage, therefore, must be converted to the appropriate percentage relative to the account corresponding to that investment style manager (5.19% in the example above). In one embodiment, back office integration server 50 converts and aggregates the trades on a per account basis and then formats the trade files as appropriate to the specific accounting package associated with account management system 80, such as Security APL. This enables the accounting/trade management system to achieve the intent of the optimization, such as picking the specific tax lots to sell (known as specific identification), ensuring that the trades match the custodial systems record of tax lots (this process is known as trading-versus-purchase), and ensuring that the appropriate purchases are executed against the cash in the account and the cash that will result from the sale of securities.
In one embodiment, optimization of a client financial portfolio involves a pre-processing phase (step 212) and an optimization phase (step 214). As discussed more fully below, in the pre-processing phase file array builder 122 and data pre-processor 124 pull required data and assemble the data arrays and files used to construct the optimization problem. The data files and arrays essentially define the optimization problem for the particular client portfolio and, in some instances, constrain the optimization process (e.g., locking a variable associated with selling a security to zero to prevent the sale of a tax lot) to ensure compliance with client objectives, portfolio construction rules, as well as tax and other regulatory issues (e.g., preventing wash sales). The optimization phase involves simultaneously optimizing the individual holdings of a client financial portfolio on an after tax basis across all investment style managers associated with and having an allocation of the portfolio.
A. Pre-Processing of Data
A.1. Process Restrictions
A restricted security is any security that a client investor has explicitly elected not to own in his portfolio. One embodiment also supports restricting categories of securities. Groups of securities that a user may restrict include sin stocks (e.g., tobacco, alcohol and gambling, etc.). Other restriction category requests may include companies “involved” in pornography, firearms, and the like.
To enforce any securities restrictions the client investor has placed on financial portfolio, data pre-processor 124, in step 312, compares any restrictions the client has placed on the account and compares them to current holdings and potential holdings (based on a review of the managers' respective model portfolios). In some cases, a client investor may have a restriction that was not violated when a security was initially purchased, but corporate actions (such as mergers and divestitures) and/or changes in the company's revenue sources may result in a violation of a security restriction (for example, a client owns a company's stock that acquires another company whose main source of revenue is from selling tobacco—the client has a restriction against holding “tobacco stocks”). When this situation exists, based on rules specified by the client (e.g., notify and request action from client, sell stock irrespective of tax consequences, etc.) data pre-processor 124 determines the appropriate action to take. In addition to current holdings, investment style managers make decisions for their respective model portfolios, irrespective of client specific restrictions. These decisions may result in the selection of a client-restricted security. In this case, data pre-processor 124 finds the most similar alternate security that meets client restrictions. By searching for alternate securities (ideally identified by the manager, but in some cases identified by other sources) and integrating these into the investment process, the client maintains a similar degree of concentration and risk/return profile for his or her portfolio. In the event that no alternate security exists and/or all alternate securities are all restricted, a hole is left in the portfolio. In other words, a model portfolio with 30 securities will now only have 29 securities during the optimization of the current portfolio. Data pre-processor 124 calculates the model portfolio weights (benchmark weights) based on the 29 securities. As discussed above, in the event that a suitable alternate security exists, data pre-processor 124 substitutes the alternate for the original model portfolio security. If the alternate type is flagged as an active risk equivalent (see Section I.C.1.a.3., supra), the alternate security gets the model portfolio weight of the original, otherwise, the model weights are calculated based on the 30 original model portfolio securities, the restricted security is included in the optimization with a zero allowed weight (UB=0) and the alternate security is included in the model but with a model portfolio weight of zero. As discussed more fully below, data pre-processor 124 enforces securities restrictions by performing such actions as modifying the security data file (e.g., flagging a security for a forced sate) or by omitting/replacing the restricted security from the array of securities in the resulting optimization problem.
If any currently held, but restricted security is locked, data pre-processor 124 keeps the restricted security in the model portfolio, otherwise, data pre-processor 124 attempts to find an alternate security. In one embodiment, data pre-processor 124 checks the model portfolios of the investment style managers associated with the client investor's portfolio for restricted securities (see step 420 and
Otherwise, data pre-processor 124 continues to the next security and repeats this sub-routine until the list of alternates is exhausted (see steps 430 and 436). If no alternate security is found, data pre-processor 124 flags the restricted security as having no alternate (step 438). Data pre-processor 124 repeats this process for all restricted securities in the current model portfolio (see step 440) and for all model portfolios of the investment style managers associated with the client's portfolio (see step 442). Data pre-processor 124 then adjusts the model portfolio as described in more detail below (see step 444 and
Beginning with the first model portfolio (see
A.2. Determine New Ideas v. Current Holdings
In pre-processing step 314, data pre-processor 124 identifies all current holdings of the client investor's portfolio and compares them to the model portfolios of the respective investment style managers associated with the current portfolio to identify any “new ideas”—i.e., any security that the client does not currently hold, but may be included in the optimization process. This process is performed separately as data pre-processor 124 performs several other pre-processing operations dependent on whether a particular security is a current holding or new idea.
A.3. Screen for Wash Sales
Per the current IRS Tax Code, a wash sale is a sale at a loss of a tax lot within 30 days before of after a purchase of the same security. The intent of the wash sale rule is to prevent a taxpayer from taking a loss on their income tax return without changing their investment position. If this rule did not exist, a taxpayer could sell a stock and buy it back immediately, thus maintaining their position but taking a tax loss on their income tax return.
In terms of a holistic investment process, wash sales are analyzed in two ways. First, based on historical client transactions outside the portfolio as they relate to past transactions within the managed portfolio. Generally the system will avoid this from happening but in the case the client provides backdated transaction data, this may occur. If this occurs, the system determines appropriate adjustments to the cost basis of all securities internal and external to the portfolios under management as more fully described in section II.A.5, infra. Second, the system screens for outside client transactions as they relate to potential transactions resulting from the current optimization of the portfolio. (See Sections II.A.3., 4. & 5.) Historical client transactions deal with any transaction executed on behalf of or made directly by the client that, when combined with all other historical transactions for that client, resulted in a wash sale. An example of a historical transaction would be when the client, in a self-directed account, buys a particular stock and, within 30 days, a manager of a wrap account owned by the client sells the same stock at a loss. The loss the manager realized would thus be passed on to the client's cost basis in the stock he purchased. The opposite may also be true where a client sells at a toss and a manager purchases the same security; in this case, the cost basis of the stock the manager purchased should be adjusted. If this is not done correctly then decisions based on the cost basis of the securities in the client's portfolio will err to the extent of the accounting transaction error.
In addition to adjustments of cost basis based on historical transactions, it is also possible to generate a wash sale based on the results of an optimization. For example, if a client sells at a loss and the manager adds the stock to his model portfolio. If this stock were purchased for the client within 30 days of the aforementioned client sale this would be a wash sale. Therefore, the optimization must be controlled to avoid this purchase and either delay the purchase, purchase a suitable alternate or conduct the purchase and account for the wash sale correctly. In the case of a sale, the optimization need only restrict the tax lots that are at a loss as wash sales do not apply to sales for a profit.
A.4. Cross-Manager Wash Sales
Due to the fact that management of a client portfolio can be allocated to at least two investment style managers, the possibility exists that an unconstrained optimization will generate a wash sale. For example, if a portion of a client's portfolio corresponding to one investment style manager holds a security at a loss and any other investment style manager adds the stock to its model portfolio, an unconstrained optimization process could possibly result in the sale of the stock at a loss and the purchase of the stock, thereby generating a wash sale. To avoid such “cross-manager” wash sales, data pre-processor 124 identifies potential wash sales and constrains the optimization to allow for either the purchase of the security or the sale of the security, but not both. In one embodiment, the optimization utilizes binary constraints to prevent such wash sales (see below). It is also important to note that, while this example describes a portion of the portfolio corresponding to one manager owning the security at a loss and another portion of the portfolio buying the security, it may in fact be possible that several portions (accounts) of the portfolio own the security and several managers desire to purchase the security, as reflected in their model portfolios.
A.5. Net Tax Position
A.6. Rebalancing, Contributions & Withdrawals
As discussed above, when client investors request withdrawals from their respective portfolios, portfolio optimization platform 40 calculates an optimal solution for achieving the withdrawal. Contributions to the client portfolio are handled in a similar manner. As the following indicates, any contribution, withdrawal and/or rebalancing decision is processed to essentially calculate the upper and lower bounds on manager and asset allocations during optimization of the current portfolio.
Contributions to a client portfolio, withdrawals from a client portfolio, and rebalancing of the client portfolio to achieve a target asset allocation are all interrelated in that they affect the allocation of the portfolio among the investment style managers and hence the overall asset allocation. Contributions and withdrawals can be targeted (the client and/or advisor selects the investment style manager to withdraw from or contribute to) or general where portfolio optimization platform 40 determines where best to route the transaction for purposes of risk, return and taxes. Similarly, rebalancing can be forced (highly discouraged) to specified targets that ignore the current tax status and estimates of risk and return for asset classes or can be optimized in which case portfolio optimization platform 40 determines the best asset allocation by considering the risk and after-tax return potential of each asset class as well as the net tax position of the client investor.
In one embodiment, file/array builder 122 retrieves and/or computes the following data for the processing of contributions, withdrawals and rebalancing (step 322): 1) contribution/withdrawal amount, 2) contribution/withdrawal target preference (assumed to be general if not specified), 3) next rebalancing date, 4) rebalance interval (e.g., in days), 5) rebalance window (e.g., in days), 6) manager allocation rebalance preferences, including deviation percentage (if any) and absolute limits (if any), 7) asset allocation rebalance preferences, including deviation percentage (if any) and absolute limits (if any), 8) all current holdings by investment style manager, 9) current market value of holdings by manager, and 10) an asset class identifier (e.g., Large Cap Growth, Mid-Cap Value, etc.) (Manager Asset Class ID) for each manager.
According to one approach, in the event that a client investor makes a contribution or withdrawal and has selected or is scheduled for rebalancing, the only option for contributions and withdrawals is non-targeted. In other words, a client cannot contribute assets targeted specifically for a manager and then have a rebalancing that ultimately moves the cash to another manager. Therefore, according to one embodiment, there are four potential cases to be implemented for any given optimization: a) targeted contribution/withdrawal without rebalancing; b) non-targeted contribution/withdrawal with or without rebalancing; c) rebalancing only (no contribution/withdrawal); and d) no rebalancing and no contribution/withdrawals.
Targeted contributions and withdrawals without rebalancing are achieved by locking all manager allocations and adding the contribution or withdrawal to the selected manager/managers. Non-targeted contributions/withdrawals are handled by adjusting the total portfolio value by the amount of the contribution (+) or withdrawal (−) and then allowing the fully invested constraint inside the optimization (optimization result as a % of the total portfolio value must equal to 100%) to force allocation of the adjustment—in this case manager allocations are slightly unlocked. Rebalancing is handled by unlocking the manager allocations and can coincide with the total portfolio value change described above. If there are no contributions and/or withdrawals and rebalancing is not scheduled or selected then manager allocations are locked to their current percentage.
A.6.a. No Rebalancing/No Contributions or Withdrawals
In the simplest case where the client portfolio is not set for rebalancing and there are no contributions or withdrawals requested (see step 904), data pre-processor 124 computes the current allocations among the investment style managers associated with the account and locks them to these percentages (step 908). In one embodiment, data pre-processor 124 calculates the current market value of each investment style manager account and sums the resulting market values to obtain the total portfolio value. Each investment style manager's allocation is the market value of the corresponding account over total portfolio value. To lock these current allocations, data pre-processor 124 sets the minimum and maximum investment style manager allocations (which are the lower and upper bounds of each managers allocation in the optimization) to the respective current allocations.
A.6.b. No Rebalancing/Non-Targeted Contribution or Withdrawal
If the client portfolio is not set for rebalancing (step 902) and a non-targeted contribution or withdrawal has been requested for the portfolio (step 904), data pre-processor 124 calculates the resulting total portfolio value (step 910) and adjusts the minimum and maximum allocations corresponding to each investment style manager based on the amount of the contribution/withdrawal (step 912). In one form, data pre-processor 124 calculates the current market value of each investment style manager account and aggregates the resulting account market values, and the amount of the contribution(s) and withdrawal(s), to obtain the total portfolio value. Data pre-processor 124 then sets the minimum allocation for each investment style manager to the corresponding account value less the withdrawal amount (if any), divided by the total portfolio value. Data pre-processor 124 also sets the maximum allocation for each investment style manager to the corresponding account value plus the contribution amount (if any), divided by the total portfolio value.
A.6.c. No Rebalancing/Targeted Contribution or Withdrawal
If a requested contribution or withdrawal is targeted for a particular investment style manager (step 906), data pre-processor 124 calculates the current market value of each investment style manager account (step 913) and adds the contribution amount to and/or subtracts the withdrawal amount from the appropriate investment style manager account(s) (step 914). Data pre-processor 124 then aggregates all investment style manager accounts to calculate the total portfolio value (step 915). Data pre-processor 124 calculates the resulting investment style manager allocations (i.e., current manager account value (after step 914)/total portfolio value) and locks these allocations (step 916). As discussed above, data pre-processor 124 locks the allocations by setting the respective minimum and maximum manager allocations to the manager allocations resulting from step 916.
If the client portfolio is set for rebalancing, data pre-processor 124 computes the values of parameters related to manager allocations and asset allocations used in the optimization to allow for rebalancing (
In addition, if the client portfolio includes asset allocation constraints (step 922), data pre-processor 124 operates to constrain the asset allocation parameters used in the optimization (step 926) (see
However, if any asset class has more than one investment style manager associated therewith and no manager allocation constraints have been computed, data pre-processor 124 associates relative manager allocation constraints to each manager account within the asset class. Relative manager allocation constraints operate to control the ratio of the amount invested between each investment style manager account within an asset class and to ensure that one investment style manager is not eliminated entirely during the optimization. In one embodiment, data pre-processor 124 determines whether each asset class includes more than one investment style manager (see steps 938, 940 and 946). If an asset class includes more than one investment style manager and no manager allocation constraints have been computed (step 942), data pre-processor 124 calculates the relative manager allocation constraints within the asset class (step 944) (see also
1) Constraining Manager Allocations
If only absolute limits are specified (step 952), data pre-processor 124 sets the minimum and maximum manager allocations to the absolute limits (step 954). Otherwise, if both deviation and absolute limits are specified (step 956), data pre-processor 124 calculates the minimum and maximum manager allocation values resulting from application of the deviation limits (step 958) (see above) and sets the minimum and maximum manager allocations values depending on whether the deviation limits exceed the absolute limits (step 960). That is, if the minimum or maximum allocation value calculated in step 958 exceeds the corresponding absolute limit, the corresponding absolute limit is chosen.
2) Constraining Asset Allocations
As described above, if only a deviation limit is specified (i.e., a maximum deviation from current asset allocations) (step 964), data pre-processor 124 calculates and sets the minimum and maximum allocation values for each asset class (step 966). In one embodiment, data pre-processor 124, ignoring contributions and withdrawals, calculates the current asset allocation values for each asset class. Data pre-processor 124 then multiplies the deviation limit by each current asset allocation value to yield the deviation amount for each asset class. To set the maximum asset allocation, data pre-processor 124 adds the resulting deviation amount for that asset class to the current asset allocation value. Similarly, to set the minimum asset allocation, data pre-processor 124 subtracts the deviation amount from the current asset allocation value.
If only absolute limits are specified (step 968), data pre-processor 124 sets the minimum and maximum asset allocation values to the absolute limits (step 970). Otherwise, if both deviation and absolute limits are specified (step 972), data pre-processor 124 calculates the minimum and maximum asset allocation values resulting from application of the deviation limits (step 974) (see above) and sets the minimum and maximum asset allocation values depending on whether the deviation limits exceed the absolute limits (step 976). That is, if the minimum or maximum allocation value calculated in step 974 exceeds the corresponding absolute limit, the corresponding absolute limit is chosen.
3) Setting Relative Manager Allocation Constraints
A.7. Construction of Data Files and Arrays
In steps 323-326, file/array builder 122 constructs the security data file 332, omega data file 334, factor data file 336, tax lot detail data file 338, manager data file 340, client data file 342, asset class return data file 344, and asset class omega data file 346, used by problem builder 126 to construct the optimization problem (see
A.7.a. Security File
In one embodiment, file/array builder 122 constructs a security file 332 comprising data fields for all securities (current holdings and new ideas) corresponding to each manager associated with the client portfolio (
1) Lower Bound (LB): LB is a parameter setting the lower bound on the security relative to the total client portfolio during an optimization. In one form, this is normally 0 but is set to the current weight of the security in the total client portfolio if the security is locked. See also Section B.3.A. & Table 5.
2) Upper Bound (UB): UB is a parameter setting the upper bound on the security relative to the total client portfolio during an optimization. In one form, this is determined by the maximum percentage of the model portfolio to invest in any one security (a portfolio construction parameter specified by the investment style manager) multiplied by the investment style manager's maximum allowable allocation of the portfolio.
3) Active Weight Lower Bound (LBawi): Active weigh lower bound is a parameter indicating the lowest value for active weight a security can take on. See also Section B.3.A. & Table 5.
4) Residual Variance (σresidual 2): This parameter is the annualized residual variance associated with the security.
5) Benchmark Weight: Benchmark weight depends on whether the tradeoffs between meeting the manager's portfolio and tax and other associated costs are explicitly or implicitly analyzed (see Section II.B.3.d, below). Under an explicit analysis, benchmark weight is the weight of the security in the benchmark portfolio the investment style manager manages against (see Section I.C.1.a.3., supra). Under an implicit analysis of the tradeoffs, benchmark weight is the weight of the security in the investment manager's model portfolio.
6) Capital Gains Alpha (αc): Capital gains alpha is the excess capital appreciation for the security forecasted by the particular manager.
7) Excess Yield Alpha (αy): Excess yield alpha is the excess yield for the security forecasted by the particular investment style manager.
8) Maximum Weight: Maximum weight is a parameter representing the maximum weight of the security relative to that portion of the portfolio allocated to the investment style manager. If the security is locked, file/array builder 122 sets maximum weight to the current weight of the security. If the current holding percentage for the security is less than the investment style manager's maximum security weight setting, file/array builder 122 sets the maximum weight parameter to the maximum security weight setting. If the security current holding percentage is greater than the maximum security weight setting, but lower than the trim level setting set by the manager, then file/array builder 122 sets the maximum weight to the trim level. Otherwise, if the current holding percentage is greater than the maximum security weight setting, file/array builder 122 sets the maximum weight parameter to the current holding percentage. Lastly, if the current holding percentage is greater than both the maximum security weight setting and the trim level setting, file/array builder 122 sets the maximum weight parameter to the trim level setting.
9) CMWS#—Cross Manager Wash Sale Number is the number assigned to the security, if any, during the screening process for cross manager wash sales (see Section II.A.4., supra).
10) Factor Exposures: Factor exposures are a series of values dependent on the risk model used by the investment style manager and the number of factors associated with that risk model. Factor exposure data can be supplied by the investment style manager, developed by portfolio optimization platform, or any other suitable source.
A.7.b. Tax Lot Detail File
File/array builder 122 then builds a tax lot detail data file 338 including parameters relating to the tax lot detail of all securities in the current holdings of the portfolio. Similar to security file 332, tax lot detail file 332 segregates tax lots by manager and security. TaxLotID is a variable the values of which are unique to and identify a particular tax lot. In one embodiment, for each manager, for each security corresponding to each manager and each tax lot corresponding to each security, file/array builder 122 builds a table comprising the following fields:
1) Transaction Cost to Buy: This parameter specifies the costs associated with purchasing the security.
2) Transaction Cost to Sell: This parameter specifies the non-tax related transactions costs associated with selling the tax lot.
3) Upper Bound of Buy (bi): Upper bound to buy is a parameter limiting the purchase of the security associated with the tax lot. In one embodiment, upper-bound-to-buy is the difference between UB (see above) and the current holdings of the security for that manager relative to the total portfolio.
4) Lower Bound of Sell (si): This reflects the least amount that must be sold of any given tax lot and is generally set to 0. See also Section II.B.3.A. & Table 5.
5) Upper Bound of Sell (si): This reflects the maximum amount permitted to sell of any given tax lot and is generally the current weight of tax lot in total client portfolio. See also Section II.B.3.A. & Table 5.
6) % Sale Taxable (Ti): Ti is the percentage of the sale of the tax lot that would be taxable.
7) Tax Lot Status (STG, STL, LTG, LTL, ULTG, ULTL): This parameter specifies the status or category of the gain or loss if the tax lot were to be sold (e.g., short term gain, long term loss, etc.).
8) Current % (ci): is the current percentage of the security across all tax lots of the security relative to the total portfolio.
9) CMWS#—Cross Manager Wash Sale Number is the number assigned to the security, if any, during the screening process for cross manager wash sales (see above).
A.7.c. Omega File
File/array builder 122 also retrieves the risk model associated with each model portfolio from market and risk model database 162 (step 326). In one embodiment, the same risk model is used across all investment style managers. In another embodiment, different risk models can be used to different investment style managers taking into consideration the particular investment style involved and/or other factors. The risk model comprises a variance-covariance matrix modeling risk in the market in relation to a set of factors. Any suitable risk model can be used, including risk models from outside sources, such as Wilshire, Barra, and Northfield. In one embodiment, during this step, file/array builder 122 places the residual volatility and factor exposures corresponding to each security in the security file 332.
A.7.d. Factor File
File/array builder 122 also retrieves the factor file 336 associated with each risk model and, as to each investment style manager, identifies the minimum and maximum factor exposures (see below). The factor file 336 contains an identification of the factors (e.g., size (log of market cap), earnings-to-price, book-to-price, net earnings revisions, etc.) included in the risk model associated with each model portfolio. In one embodiment, the factor file 336, as to each manager, contains a numbered list of the factors. As to each factor, the factor file 336 includes the following fields:
1) Factor Name: Factor name is a text variable indicating the name for the factor.
2) Factor Abbreviation: Factor Abbreviation is a text variable representing an abbreviated form of the factor name used to facilitate generation of reports and the like.
3) LB Fj: Minimum factor exposure is the smallest factor exposure value for that factor from the set of securities in security file 332 corresponding to the investment style manager.
4) UB Fj: Maximum factor exposure is the smallest factor exposure value for that factor from the set of securities in security file 332 corresponding to the investment style manager.
A.7.e. Asset Class Return Data
File/array builder 122 also retrieves a set of forecasts for the expected return of each asset class to which the client has exposure. In one embodiment, this forecast is developed by personnel managing the system. In another embodiment, the forecast is provided by the distributor of product enabled by the system. As to each asset class the file contains the following fields:
1) Asset Class Name: Name is a text variable that refers to the asset class such as All Cap Core.
2) Return (Asset Class): A forecast of the expected percent return of the asset class over a 1 year time horizon (such as 12%).
3) Lower Bound of Asset Class: Reflects the minimum amount post-optimization the asset class should represent of the total client portfolio.
4) Upper Bound of Asset Class: Reflects the maximum amount post-optimization the asset class should represent of the total client portfolio.
A.7.f. Asset Class Omega Data
File/array builder 122 also retrieves a variance/covariance matrix that describes the interrelationship among asset classes for which the client has exposure. In one embodiment, this variance/covariance matrix is developed by personnel managing the system. In another embodiment, variance/covariance matrix is provided by distributor of product(s) enabled by the system.
A.8. Manager and Client Files
File/array builder 122 then builds manager and client data files, in part, based on the results of the pre-processing steps discussed above, as well as settings and preferences stored in client account database 164 and investment style manager database 151.
A.8.a. Manager Files
File/array builder 122 also builds a manager data file 340 including data for each investment style manager associated with the client portfolio. In one form, manager data file 340, for each manager, includes the following fields:
1) ManagerID: ManagerID is a parameter identifying the particular investment style manager. A particular manager may be associated with more than one ManagerID value, in cases where the investment style manager manages more than one investment style.
2) Asset Class ID: Asset Class ID is a parameter identifying the particular asset class associated with the investment style manager.
3) Number of Factors in Risk Model (J): This parameter indicates the number of factors in the risk model associated with the investment style manager.
4) Number of New Ideas+Current Holdings (N): This parameter indicates the number of new ideas identified in pre-processing step 314 for each investment style manager plus the current holdings specified below.
5) Number of Current Holdings (X): This parameter provides the number of current holdings in that portion of the client portfolio allocated to the investment style manager.
6) Number of Tax Lots (TL): This parameter indicates the number of tax lots in that portion of the client portfolio allocated to the investment style manager.
7) LB Capital Gains Alpha (αc): This parameter is the minimum capital appreciation alpha among the securities associated with the investment style manager in the security file 332. This parameter, as well as Max αc, Min αy, and Max αy, are used as upper and lower bounds in the optimization. See below.
8) UB Capital Gains Alpha (αc): This parameter is the maximum capital appreciation alpha among the securities associated with the investment style manager in the security file 332.
9) LB Yield Alpha (αy): This parameter is the minimum yield alpha among the securities associated with the investment style manager in the security file 332.
10) UB Yield Alpha (αy): This parameter is the maximum yield alpha among the securities associated with the investment style manager in the security file 332.
11) (λS): This parameter quantifies the client investor's tolerance level for systematic risk.
12) (λU):This parameter quantifies the client investor's tolerance level for unsystematic risk.
13) LB Mk: This parameter specifies the minimum allowable allocation of the client portfolio corresponding to the investment style manager. When optimization does not involve rebalancing, investment style manager allocations are locked; therefore, the minimum and maximum allocations (below) are set to the current allocation. In a rebalancing period, however, these parameters are loosened to allow for a range of allocations per investment style manager. If a contribution or withdrawal targeted at a specific investment style manager is indicated with no rebalancing, the current allocation percentages are adjusted accordingly and locked as above. (See Section II.A.7., supra).
14) UB Mk: This parameter specifies the maximum allowable allocation of the client portfolio corresponding to an investment style manager resulting from an optimization. As discussed above, this may be set to the current allocation. (See Section II.A.7., supra).
15) Relative Weight ID (RWID): RWID is a unique identifier for the manager, if there is more than one manager in the corresponding asset class.
16) Relative Weight Minimum: In the case where there is more than one manager in an asset class, this parameter sets the lower bound on the relative weight within the asset class for the manager.
17) Relative Weight Maximum: This parameter sets the upper bound on the relative weight for the manager within the asset class, if the asset class includes more than one manager.
A.8.b. Client File
File/array builder 122 also retrieves data from client account database 164 and builds client data file 342. Client data file 342 includes client-specific data (such as applicable tax rates), as well as the client's preferences and other settings influencing optimization of the user's portfolio. In one embodiment, client data file 342 includes the following fields:
1) Income Tax Rate: This field includes data relating to the income tax rates (federal and state) applicable to the client investor.
2) Capital Gains Tax Rate: This field includes data relating to the capital gains tax rates applicable to the client investor. Because this is not client-specific, such data could be pulled from general tax rate tables.
3) Lambda(O): Lambda(O) is an asset class tolerance parameter identifying the client's tolerance for asset class risk.
4) Lambda(T): This is a tax tolerance factor that takes into account the time value of money for tax dollars that don't have to be paid until some point in the future. The implementation of this factor reflects the percentage of the applicable tax rate that applies to compensate for this effect. In optimization, the tax tolerance factor is multiplied by the tax rate to come up with an effective tax rate.
5) Number of Asset Classes (A): This parameter indicates the number of unique asset classes associated with the client portfolio.
6) Number of Multi-Manager Asset Classes (Q): This parameter indicates the number of managers in asset classes having more than one manager.
7) Number of Managers (M): This is a parameter indicating the number of investment style managers associated with the client investor's portfolio.
8) Number of Cross Manager Wash Sale Restrictions (H): This is a parameter representing the number of cross-manager wash sale restrictions identified in a pre-processing step (see above).
9) Total Portfolio Value (TPV): Total Portfolio Value is a parameter indicating the total value of the client portfolio.
10) Tax Constraints:
As described more fully below, the optimization technology optimizes the expected return of a financial portfolio allocated to a plurality of investment style managers from the perspective of the client, as opposed to a single money manager utilizing an optimization platform for a portfolio (with or without respect to taxes) in isolation. A typical optimization product that can deal with tax lot level data is not designed to deal with the same security existing in multiple accounts, as may occur during operation of the present invention. First, this circumstance introduces the complexity of treating each portion of the client portfolio associated with a different investment style manager (accounts) separately while still considering the aggregate correlation amongst accounts. Second, the assumption of an optimizer in isolation is that the single portfolio is the single investment for a taxpayer. In reality, a taxpayer may have multiple accounts that should be considered to appropriately allocate assets and tax losses amongst the accounts. Third, the introduction of multiple managers means that what would otherwise be simple constraints for optimization, such as the maximum amount to invest in a given security become significantly more complex algebraically. For example, if a specific account for a client will not allow more than 5% to be invested in any single security, this translates to 2.5% if the manager's account comprised 50% of the total portfolio. Unfortunately, since the manager allocation may change during a rebalancing period (a shift of assets amongst asset classes and hence individual managers) as well as when a contribution or withdrawal is made, the manager's allocation of the total clients assets is not necessarily fixed. Therefore many constraints must be algebraically adjusted in relation to the respective allocations among investment style managers for each client portfolio (see below).
In one embodiment, upon completion of the required files and arrays (see above), process handler 120 passes execution to optimization problem builder 126 which operates to construct the optimization problem. An exemplary optimization problem is set forth below (see Section B.3., below). For didactic purposes, however, the following sets forth certain optimization concepts and the modifications required to implement the present invention.
B.1. Effects of Simultaneous Optimization involving Multiple Managers
One example of how investment style manager allocations affect the optimization process is when an account must have a factor weighting less than, greater than or equal to a specific amount. For example, if a specific account must have an average price-to-earnings ration (P/E) of less than 12. Software that deal with this constraint in isolation to the total portfolio would require that, for n securities in the account where each has a weight of wi and Price to Earnings of (P/E)i, the account would have the following constraint.
For didactic purposes, assume that this account is currently 50% of the client's total managed portfolio. One of two choices could be made in the formulation of such a problem, the weights (wi) could sum to 100% for each account or the weights (wi) could sum to 100% relative to the entire portfolio. If the weights sum to 100% for each account, the constraint expressed in equation 1 above would still hold. However, there would be no distinction between two accounts of differing allocations, thereby creating complications when trying to determine allocations amongst managers and when determining client portfolio level variables that are a function of individual tax lots within a manager (such as the net tax effect to the client). In order to deal with this, the optimization must therefore have the security weights across all accounts in the portfolio sum to 100. In this instance, an individual account's weights aggregate to less than 100%. Accordingly, for an account comprising 50% of the total portfolio, the constraint above set forth in equation (1) becomes
As mentioned above, however, the 50% allocation is not a fixed allocation to that manager. If the allocation corresponding to this account is the variable M, the correct representation of this account constraint then becomes:
for all securities under consideration or held by that account.
Since optimization is represented with all variables on one side of the equation, these equations are combined algebraically to correctly configure the optimization constraint for the account. For example, for didactic purposes, assume that there are three stocks under consideration by an investment style manager with weights of x1, x2, x3 and another investment style manager involved in the optimization has two stocks with weights x4 and x5. It follows that the average P/E for stocks corresponding to the first manager (with P/Es of 10.5, 12.5 and 14.2) is:
Equation 4 must be simplified for use in the optimization problem solver. With the right hand side of the equation set to 12 (the average P/E desired among the securities), multiplying the left hand side of the equation through by the denominator yields:
10.5x 1+12.5x 2+14.2x 3+0x 4+0x 5=12(x 1 +x 2 +x 3). (5)
However, in order to format equation 5 for a solver, all terms are brought back to the left hand side. This can be done by first multiplying through by 12, as shown below:
Then, subtracting the terms from the right hand side from both sides of the equation:
Finally, combining the terms to collapse all calculations into single coefficients yields:
So, in generalized form, the left hand side of the constraint in a partial factor loading is:
Where f1, f2 and f3 are the data elements for the particular factor (such as P/E) and T is the target value for the account. The term in front of each variable represents the deviation from the target value. In the case of an inequality constraint (the average of the three must be less than or equal to 12), the equals sign simply becomes:
Equation 10 represents a constraint requiring that the weighted deviation for those three securities from the average of 12 must be less than 0, therefore the average must be less than 12.
Most constraints in the optimization can be generalized as either a factor loading constraint (the weighted average of a factor and the amount invested in each security within that account) or a summation constraint (the amount invested amongst a group of securities should be greater than, less than or equal to some amount). Therefore many of the constraints must be adjusted to reflect the allocation of the manager in a similar fashion.
B.2. Simultaneous Optimization
As discussed above, each investment style manager has an allocation of the overall client portfolio (also referred to as an account in the example above). The manager's portion of the total portfolio comprises an array of financial assets. Each manager, over time, will introduce new assets for consideration and sell other assets represented by changes in the model portfolio submitted to portfolio optimization platform 40. When this occurs, a tradeoff analysis needs to be conducted by analyzing what is the benefit of the trade versus the cost to achieve the trade. The optimization process, therefore, considers the tradeoff between the cost to trade (transaction costs), the difference in the return opportunity and the difference in the risk. Since trades may (in a taxable account) have a tax cost associated with them, the cost of the trade must be considered in terms of its taxes. This tax cost, however, is not unique to a given account but is a function of the activity in all of the accounts. It follows that, tax costs should be considered in the aggregate, as opposed to at the individual stock level. Additionally, most managers are attempting to outperform a benchmark and deal with the decision to buy and sell relative to some establish benchmark that they are trying to outperform. Therefore, while the manager's information helps to determine which trades to make within a manager account, it does not help to determine how best to allocate securities across accounts based on the clients overall tolerance for risk, return and tax costs. Therefore, a simultaneous optimization requires inputs at a “manager level” and at a “client portfolio level” that support making decisions within an account (what to buy and sell) and on behalf of a client (how to allocate resources and where to realize tax gains and losses). While there are two distinct levels of consideration they are interrelated; as explained earlier, the only way to affect the allocation amongst managers is by trading, which has implications for purposes of taxes.
B.2.a. Overview of Optimization Generally
For didactic purposes, a general overview of the processes associated with optimization is provided. Optimization generally involves the process of representing the tradeoffs described above mathematically and then using algorithms designed to minimize or maximize mathematical relationships. Optimization typically has four components that describe the optimization: 1) the objective function (i.e., the mathematical relationship of tradeoffs) to be maximized or minimized; 2) a list of variables, including the number and type (continuous, binary and integer); 3) the bounds on variables; and 4) constraints. As described earlier, in the impact of manager allocations on optimization calculations (see Section II.B.1., supra), many of the constraints require additional algebraic representation to handle the multiple-manager paradigm.
Optimization of linear, quadratic and integer variables can be accomplished via software products that specialize in implementation of algorithms for solving various mathematical problems. The following describes how the optimization of client portfolios translates to standard optimization software packages. Different algorithms are utilized by different software vendors based on the nature of the problem and the goal of the optimization (such as global optimization or to find a feasible solution). The inputs to optimization software products include settings on how to solve the optimization (number of iterations, selection of algorithm, etc.) and the problem itself. Any optimization problem comprises a set of variables. While algebraically, we may represent a problem with meaningful variable names and in a meaningful orientation within the equation, this is meaningless to the actual optimization software which merely refers to all variables as x1, x2, x3, etc. Therefore, to specify a problem for solution, the inputs to the software include: 1) the total number of variables, 2) variable types (continuous, integer or binary), 3) the range of possible values for the variables (upper and lower bounds), 4) constraints (relationships to variables on the left hand side, an equality or inequality and right hand side values), and, 5) in the case of quadratic optimization, the quadratic relationship between any two variables. See
In order to understand the optimization, one must first conceptualize the variables in the equation as inputs in a specific order. All other inputs must be aligned with the variables in a similar fashion.
1) Variable Types
All variables in the problem can be either continuous, integer or binary (a special case of integer). In the present optimization, most variables are continuous, however, binary variables are utilized to constrain the optimization from wash sales, prevent simultaneous transactions (selling and immediately buying back) and to accurately model the US tax code. Note in this subset model, variables o1 and o2 are binary.
2) Variable Bounds
Each variable in the optimization can be bound to some range. For instance, for w2, the maximum value under any circumstances that could be invested in security 2 (as a percentage of the client investor's portfolio) is 30% in this example.
B.2.c. Linear Objective
The objective function is the set of variables and relationships that are being maximized and/or minimized. Anything that is in the linear portion of the objective function appears in the objective array with a coefficient that represents the value that variable has in the problem. For example, in the simultaneous optimization, trades are determined for each manager/account that determine the optimal tradeoff between return, risk and transaction costs. The linear terms in the objective function are return and transaction costs. Note that the objective array, therefore, has transaction costs associated with the b and s variables (variables representing the buying or selling of a security, respectively) and that the variables having a return associated with them (e.g., αc, αy, and fj) have coefficients that represent the amount of return attributed to a unit change in that variable. It is also important to note that the alpha variables (αc, αy) are linked via a constraint to the security weights (see constraints section).
B.2.d. Quadratic Objective
The quadratic objective is the portion of the objective function comprised of the relationship between two variables. In the invention, all risk variables are quadratic Since quadratic relationships occur between any two variables, the quadratic portion of the objective function is (for N variables) an N×N array of the coefficients that describe the relationship. As illustrated in
−λsσs 2, (11)
where the variance components are represented by the following mathematical relationship:
When the inputs are provided as shown above, the negative lambda term is distributed amongst the omega matrix to accomplish the objective function input from equation 11. As a specific example from equation 12 where j=1 and k=2, the objective function result is:
This is achieved by the shaded outlined box in the quadratic matrix as are all similar instances of j and k in equation 12 and all other quadratic terms.
Constraints can be utilized for several functions in an optimization problem. The constraints shown in the sample problem above illustrate constraints that are utilized for three general purposes: 1) creating a variable that is a function of several other variables (composite variable constraint); 2) enforcing a logical condition that must be met (binary logic constraint); and 3) constraining a set of variables in aggregate from exceeding some bound (summation constraints). While these three examples cover many of the constraints used in portfolio optimization, constraints can create/enforce any mathematical/logical relationship between variables. Each of the three examples is discussed in greater detail in the following sections.
1) Composite Variable Constraint
Composite variable constraints are used to express factor loading variables. In the example above, the factor loading variables (f1, f2, f3) are the weighted average factor exposures for a portfolio account. In mathematical terms, a factor loading would be represented as:
In the example provided by
f j =f j,1 w 1 +f j,2 w 2 +f j,3 w 3 +f j,4 w 4 +f j,5 w 5 (15)
With all non-constant terms to the left hand side, equation 15 becomes:
f j,1 w 1 +f j,2 w 2 +f j,3 w 3 +f j,4 w 4 +f j,5 w 5 −f 1=0 (16)
2) Binary Logic Constraint
Another type of constraint utilized in the example above is a binary logic constraint. Integer variables can be utilized for a multitude of purposes in financial optimization such as ensuring that the weight of a variable equates to a trade in increments of 100 shares (round lot constraints), limiting the number of securities in the portfolio to a certain number (cardinality constraints) and limiting a new security entry unless they are of some minimum size (buy-in constraints). Binary constraints are a special case use of integer values where the value is either 1 or 0. Binary constraints support the last two purposes described above and can also serve to enforce some logical condition.
The use of binary constraints for purposes of logic allow the creation of either/or and if conditions. In one form, binary constraints create “buy-or-sell-only” constraints used to prevent wash sales. In the example of
where bi represents a purchase of additional amounts of security i, si,x represents the sale of some or all of tax lot x of security i and oi represents a binary logic variable. It should be noted that, if the resulting value of b1 is any amount other than zero (the optimization recommends buying more of a specific security), then, in order to make equation 17 true, o1 must be equal to 1 (the lower bounds on b and s are zero). In addition, if o1 equals 1, both s variables in equation 18 must be equal to zero to keep the second equation true. Similarly, where either or both s variables are greater than 0 (representing the sale of one or both tax lots), equation 17 requires that b1 must therefore be equal to zero. The combination of these two constraints, therefore, forces only one of two logical conditions to exist.
3) Summation Constraints
Summation constraints can impose a limit on a given combination of variables. A good example of this type of constraint is a sector weighting constraint. A summation constraint is a more specific case of a factor loading constraints where all factor values are equal to one. The equation for a factor loading as described in the previous section was:
The translation to left hand and right hand side elements is identical for a summation constraint and the only differences is that instead of data elements fi,j, no value for the coefficients (equivalent to 1) is used. Therefore the equation for a summation constraint becomes:
In the case of the sum of some group of variables, only the variables that meet some condition (such as in a sector, in a manager's sub-account, etc.) receive a coefficient of 1. All other variables in the entire optimization problem will have a coefficient of 0 (the equivalent of dropping them from the mathematical relationship).
B.3. Exemplary Optimization Problem
The following sets forth an exemplary optimization problem according to an embodiment of the present invention. The following sets forth variables, types, and bounds, as well as an objective function and various optimization constraints.
B.3.a. Variables, Types and Bounds
1) Manager Specific Variables
Table 4 sets forth manager-specific variables and variable types according to one embodiment of the invention. The values of the variables are specific to each investment style manager.
Weight of each security i as a % of total portfolio
Active weight of each security from benchmark
(custom or index)
Active factor exposure from benchmark (custom
% To buy for current holdings as a percentage of
% To sell of each tax lot as a percentage of
total portfolio value
Buy or Sell Only binary variable
Expected excess return attributable to capital
Expected excess return attributable to dividend yield
In addition, in one embodiment, the optimization includes the following variable bounds for the manager specific variables identified above.
As discussed above, in the pre-processing step, data pre-processor 124 and file/array builder 122 construct the files and arrays containing the data, including the upper and tower bounds on variables, required to construct the optimization problem. Table 5 sets forth the upper and lower bounds for certain manager-specific variables. Note: MPS represents maximum position size, a parameter indicating the maximum weight of the security as if the manager's sub-account was the only account being optimized. This maximum size is directly tied to the model portfolio weight of the client specific model. For example, a manager's model portfolio has a weight for a security of 2%. The maximum position size would therefore be a function of 2%, such as 2×2% of 4%. Generally, the maximum position size is set to a multiple of the model portfolio weight.
Current Percentage of security in total
Current Percentage of security in total
MPS times Maximum Manager Allocation
<MPS, Buy More
Current Percentage of security in total
Current Percentage of security in total
Trim to Level times Manager Maximum
Negative Benchmark Weight of security
Minimum Factor Exposure Value from
Security File for this Manager
Maximum Factor Exposure Value from
Security File for this Manager
Max (O, UBw
security in total client portfolio)
Weight of current lot in total client
Weight of current lot in total client
Minimum αc Value from Security File for
Maximum αc Value from Security File for
Minimum αγ Value from Security File for
Maximum αγ Value from Security File for
2) Client-Specific Variables
Table 6 sets forth client-specific variables and variable types according to one embodiment of the invention.
Cross Manager Wash Sale Variables
Short Term Losses as a percentage of total
Short Term Gains as a percentage of total
Long Term Losses as a percentage of total
Long Term Gains as a percentage of total
Net Short Term as a percentage of total portfolio
Net Long Term as a percentage of total portfolio
Net Gain or Loss as a percentage of total portfolio
Binary Logic Variables to Aid in Calculation
Taxable Gains as a Percentage of Total Portfolio
Value for Certain Tax Situations
Manager Allocation Variables
Asset Allocation Variables
In one embodiment, the bounds on the client-specific variables above include:
Table 7 illustrates the potential lower bounds (LB) and upper bounds (UB) values for client specific variables in one embodiment of the invention. As table 7 shows, the upper and lower bounds values may be influenced by client-specific inputs or preferences concerning taxes, such as a preference for avoiding aggregate short term gains, avoiding short term gains absolutely, etc.
1 + (External ST Losses/Total Portfolio
(External ST Losses + Forced Sell ST
Losses)/Total Portfolio Value
1 + (External ST Gains/Total Portfolio
(External ST Gains+ Forced Sell ST
Gains)/Total Portfolio Value
1 + (External LT Losses/Total Portfolio
(External LT Losses + Forced Sell
LT Losses)/Total Portfolio Value
1 + (External LT Gains/Total Portfolio
(External Long Term Gains + Forced
Sell Long Term Gains)/Total Portfolio Value
1 + (Sum External Gains and Losses/Total
Max [((Sum External + Sum Forced − Sum
All Potential Losses)/Total Portfolio
1 + (Sum of External ST Gains and
Losses/Total Portfolio Value)
Max [((Sum External ST + Sum Forced
ST − Sum All Potential ST)/Total Portfolio
1 + (Sum of External LT/Total Portfolio
Max [((Sum External LT + Sum Forced
LT − Sum All Potential LT)/Total Portfolio
Minimum Manager Allocation From Con-
tribution, Withdrawal and Rebalancing
Maximum Manager Allocation From Con-
tribution, Withdrawal and Rebalancing
Minimum Asset Class Allocation From Con-
tribution, Withdrawal and Rebalancing
Maximum Asset Class Allocation From Con-
tribution, Withdrawal and Rebalancing
B.3.b. The Objective Function
The goal of the optimization is to maximize the investor's utility where utility is represented as a relationship among the preferences for return, risk and taxes across one or more accounts that make up the client investor's portfolio. In one embodiment, utility is mathematically represented by the objective function set forth below. The objective function can be split into the linear and quadratic portions as represented below:
Client's Marginal Income Tax Rate
Client's Marginal Capital Gains Tax Rate
rIT − r1
Transaction costs associated with a unit percent change of
Note: αc adjusted for transaction cost for securities not in
Asset Class Expected Return
Systematic Risk Tolerance Parameter for Manager m
Stock Specific Risk Tolerance Parameter for Manager m
Variance/Covariance Term for Systematic Risk
Residual Variance Term for Security i, Manager m
Overall Risk Tolerance Parameter for Asset Allocation
Variance/Covariance Terms for Asset Class Risk
Constraints can be broken down into both client and manager level (although some client level are linked to the manager level) and further broken down into the purpose for the constraint. The following sets forth exemplary constraints as well as the pre-processing steps required for each constraint.
1) Client Level Constraints
a) Tax Variable Constraints
The tax variable constraints are a series of four constraints [c1, c2, c3, c4] that sum the total taxable events in each tax status category for each tax lot where there are Z total tax lots in the entire client portfolio. Ti reflects the taxable portion of any sale of a unit of a tax lot.
Pre-Processing Data Requirement
For each tax lot, data pre-processor 124 calculates Ti per the following conditions/formulas:
Short Term or Long Term Gain:
Short Term or Long Term Loss:
Where C=the cost basis of the security, and V is the market price of the security.
In addition, for each tax tot, data pre-processor 124 identifies its tax status and code as follows:
Short Term Loss→LotTaxStatus=0
Short Term Gain→LotTaxStatus=1
Long Term Loss→LotTaxStatus=2
Long Term Gain→LotTaxStatus=3
Furthermore, data pre-processor 124, calculates external taxable gains and losses as a percentage of the total portfolio value as described in the pre-processing step 320 (see Section II.A.5., supra).
b) Tax Logic Constraints
The tax logic constraints are a series of constraints that accurately model the then-applicable US Tax Code. These constraints (set forth below) result in one of four potential states:
c) Manager Allocation Constraints
Manager allocation constraints constrain the total amount invested in a given manager account as a percentage of the total client portfolio. The number of managers is reflected in k variables, one for each manager. The constraint is calculated per manager for securities in the managers sub-account per the following formula:
Pre-Processing Data Requirement
All securities are sorted in the security file by manager. The Manager file contains a count of all securities in the file. During the building of the problem, the securities for each manager are processed and a unique constraint is built in the client constraints section (i.e., on a unique row) that is linked to the Mk variable in the client variable section.
d) Manager Relative Allocation Constraint
The manager relative allocation constraint ensures that the ratio of assets among managers within an asset class is bounded by some range. For each manager in the same asset class, two (minimum relative, maximum relative) constraints are created per the following formulas:
Data pre-processor 124 generates a count (Q) of the total number of managers that require manager relative allocation constraint (see Section II.B.7., supra). For each manager that requires relative asset class allocations, the minimum and maximum should be calculated. In one embodiment, managers are bounded between their original allocation at account funding and their current allocation. These calculations are performed in the Contribution, Rebalancing and Withdrawal calculations process (see Section A.6., supra).
d) Asset Allocation Constraint
For each asset class (L) and all managers (m) within the same asset class ID (L), optimization problem builder 126 builds one constraint per asset class to reflect the total client assets as a percentage of the total client portfolio invested in each asset class. There should be L total constraints.
Note: During pre-processing, an asset class ID was placed in the manager file for each manager associated with the current client portfolio.
e) Fully Invested Constraint
The fully-invested constraint ensures that the clients assets in the managed portfolio are fully invested among the potential asset classes. One constraint is built for the client to ensure that the investment into all (L) asset classes totals the total client portfolio value 1.
f) Cross Manager Wash Sale Constraints
Cross manager wash sale constraints ensure that, during any optimization involving more than one manager, no two managers effectively generate a wash sale. The most likely time this would ever occur is when two managers within the same asset class have a different opinion on any single security. If one manager removes the security from his model portfolio and another adds the security to his respective model portfolio, depending on the state of the gain or loss, a potential wash sale could occur. This series of constraints ensures that the most beneficial trade is achieved. The other manager either postpones the trade and/or purchases an alternate security. In order to require this constraint, at least one manager must hold a tax lot at a loss and any other manager may hold or consider buying this security. Both current holdings (denoted by bi and si variables) and potential holdings (denoted by wi variables where the current percent held of security i [denoted by ci] equals zero).
ci=current percentage invested in security i,
CMWS=a logical condition set to true if a security has been flagged as a potential cross manager wash sale;
X=count of all tax lots within any manager;
M=total number of managers; and
N=total securities held or being considered within a manager's sub-account.
Pre-Processing Data Requirement
As discussed above, as a pre-processing step, data pre-processor 124 identifies cross manager wash sales process. Each security is screened for this potential condition and all securities that qualify are labeled with an associated cross manager wash sale ID. When a security does not qualify as a potential cross manager wash sale the ID is set to zero. In addition, the client file includes a count of the total number of cross manager wash sale potentials within the entire optimization.
2) Manager Level Constraints
a) Active Security Weight Constraints
Active security weights calculate the portion of an investment in a given security that represents active risk. Active risk is deviation from a specified benchmark (either an index or a private portfolio such as a manager's model). For a discussion on what benchmark is used see the implementation section below. Since the benchmark weight is represented as if the client has invested 100% of their assets in the benchmark, the formula must be adjusted to reflect the percentage allocated to the manager account. A constraint is created for each security under consideration or currently held by a manager per the following formula:
Where bwi=the benchmark weight for the specific security.
Pre-Processing Data Requirement
During pre-processing, the client specific model portfolio must be constructed. In the case of implicit analysis (to be discussed later) the client specific model portfolio is the benchmark and in the case of explicit analysis, an asset class passive benchmark is the established benchmark. Each securities benchmark weight must be appropriately identified and stored in the security file.
b) Individual Security Weight Constraints
The individual security weight constraint ensures that no more than a specified amount is invested in any given security. In most instances, there are not many occasions where this constraint is forced to be implemented since the client will more than likely deviate substantially from a benchmark weight and/or model portfolio weight and thus the position will be trimmed versus exceeding some max weight limit. Nonetheless, the system supports setting some maximum weight on a case-by-case basis for each security/manager/distributor/client. Since maximum weights are typically referred to relative to a single manager account (i.e., a manager trims any security that exceeds 8% of the portfolio back to 4%), they must be adjusted for the current allocation the manager account has in the total client portfolio. The constraint can be represented mathematically for each security as:
wmi=the maximum weight to invest in security i;
wi=the weight as a percentage of total portfolio of security i; and
wk=the weights of all other securities under consideration for that manager.
Pre-Processing Data Requirement
Max weights are calculated during the building of the security file per the following conditions table:
Current Percentage of security in total client
portfolio/Current Manager Allocation
MPS (i.e. the maximum position size for the
<MPS, Buy More
Current Percentage of security in total client
portfolio/Current Manager Allocation
Current Percentage of security in total client
portfolio/Current Manager Allocation
Trim to Level/Current Manager Allocation
c) Return Vector Constraints
The return vector constraints collapse the forecasted return/excess return for each manager into two variables attributed to capital appreciation and yield. Those skilled in the art might recognize that it would be possible to include these alphas in the objective function for each security weight (wi), however, the security alphas would have to be adjusted for the client's specific income tax rate. Additionally, the output of the optimization in this fashion gives an explicit output of the expected return/excess return attributed to capital gains and yield.
Pre-Processing Data Requirement
As discussed above, alphas and other model portfolio data are received from investment style managers on a periodic basis and stored in investment style database 150 and processed by investment style data processor 154 (see Section I.C.1.a., supra). These calculations convert the input to a normalized alpha forecast in excess of the benchmark return. The normalized alphas are further broken down into excess return attributable to capital appreciation versus yield. In order to achieve this breakdown, the yield of the security is subtracted from the expected yield of a benchmark to determine the yield alpha. The yield alpha is subtracted from the total alpha to determine the alpha attributable to capital appreciation. It is feasible based on these calculations to have a negative yield alpha and/or a negative capital appreciation alpha. When a specific client account is run, the alphas are pulled from investment style database 150 and stored in the security file. In one embodiment, for securities not in the current holdings, the expected transaction costs as a percentage of the security price is subtracted from the capital appreciation alpha.
d) Active Factor Loadings Constraints
Active factor loadings reflect the amount of active systematic risk from a specified benchmark. In the case of the present invention, the factor exposure data is calculated and stored in the security file during pre-processing and one constraint per each (J) factor exposures is created for each manager/sub-account.
N=total securities held or under consideration by manager;
BFj=the benchmark exposure to factor j; and
fei,j=security i's factor exposure to factor j.
Pre-Processing Data Requirement
In the pre-processing phase, data pre-processor 124 calculates the difference between the benchmark factor exposure and the stock specific factor exposure and stores this in the securities file for each security within each manager.
e) Current Holdings Constraints
This constraint links the tax lot detail for each specific security to the total weight invested in the security. The total weight includes how much is held before the start of the optimization (ci), how much is purchased (bi) and how much is sold of each tax lot (si,m where there are y specific lots for any security i).
Pre-Processing Data Requirement
For this constraint, no pre-processing is required. It is critical that all securities are labeled properly such that the security and tax lot files correctly identify the same security so that this constraint can be built correctly.
f) Buy or Sell Only Binary Constraints
This constraint ensures that for the specific tax lots currently held in any specific sub-account that either more is purchased or specific lots are sold, but not both. This eliminates the opportunity to create a wash sale during the optimization. If the manager has a relatively high alpha forecast compared to the marginal risk of this security in the total portfolio then it would be beneficial to buy more. If the optimization were left unconstrained then the optimization would allow selling all lots at a loss and reinvesting all proceeds back into the same securities thus giving an unrealistic picture of the tax state for this investor as this would result in a wash sale. Note that wash sales based on previous transactions (either purchases within the last 30 days and a sale at a loss OR a sale within the last 30 days and a purchase during the optimization) are constrained by locking specific tax lots and security weights in pre-processing. The mathematical relationship for each security i is as follows:
Pre-Processing Data Requirement
For this constraint, no pre-processing is required. It is critical that all securities are labeled properly such that the security and tax lot files correctly identify the same security so that this constraint can be built correctly.
B.3.d. Configuration of Optimization
As described in Section I.C., supra, there are two methods for making the tradeoff analysis in a manager's portion of the portfolio, explicit and implicit analysis. In the case of explicit analysis, the manager provides a model portfolio including explicit rankings of the securities within the model portfolio. Investment style data processor 154 converts the explicit rankings into capital appreciation and yield alpha values for each security. The benchmark is an index (e.g., Russell 1000 Growth) against which the investment style manager manages. λs and λu reflect the tradeoff between tracking error and alpha. Transaction costs (tci) are a function of the bid/ask spread for the security and any estimated market impact (current assumption is 0). The factor exposures, factor variances, factor correlations and residual variance forecasts can be from any commercially available risk model provider.
In the case of implicit analysis, all capital appreciation and yield alphas are set to zero. The benchmark is the manager's model portfolio. λs and λu reflect the tolerance for deviating from the model for purposes of taxes. tci is a function of the bid/ask spread for the security and any estimated market impact (current assumption is 0). The factor exposures, factor variances, factor correlations and residual variance forecasts can be from any commercially available risk model provider. In addition, as discussed above, the problem is constructed and solved to yield a set of recommended trades to achieve the results of the optimization. (See Section II, supra).
Lastly, the present invention has been described with reference to specific embodiments. Other embodiments of the present invention will be apparent to one of ordinary skill in the art. It is, therefore, intended that the claims set forth below not be limited to the embodiments described above.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5819238||Dec 13, 1996||Oct 6, 1998||Enhanced Investment Technologies, Inc.||Apparatus and accompanying methods for automatically modifying a financial portfolio through dynamic re-weighting based on a non-constant function of current capitalization weights|
|US5819263||Jul 19, 1996||Oct 6, 1998||American Express Financial Corporation||Financial planning system incorporating relationship and group management|
|US5918217||Dec 10, 1997||Jun 29, 1999||Financial Engines, Inc.||User interface for a financial advisory system|
|US5930774||Jan 29, 1996||Jul 27, 1999||Overlap, Inc.||Method and computer program for evaluating mutual fund portfolios|
|US5978778||Dec 20, 1997||Nov 2, 1999||O'shaughnessy; James P.||Automated strategies for investment management|
|US5987433||Jun 3, 1996||Nov 16, 1999||General Electric Company||Method and system for developing a time horizon based investment strategy|
|US6012043||Sep 9, 1996||Jan 4, 2000||Nationwide Mutual Insurance Co.||Computerized system and method used in financial planning|
|US6012044||May 25, 1999||Jan 4, 2000||Financial Engines, Inc.||User interface for a financial advisory system|
|US6018722||Jun 19, 1997||Jan 25, 2000||Aexpert Advisory, Inc.||S.E.C. registered individual account investment advisor expert system|
|US6021397||Dec 2, 1997||Feb 1, 2000||Financial Engines, Inc.||Financial advisory system|
|US6052673||Nov 2, 1998||Apr 18, 2000||Trans Texas Holdings Corporation||Investment management|
|US6064985||Jan 21, 1998||May 16, 2000||Assured Equities, Inc.||Automated portfolio management system with internet datafeed|
|US6078904||Mar 16, 1998||Jun 20, 2000||Saddle Peak Systems||Risk direct asset allocation and risk resolved CAPM for optimally allocating investment assets in an investment portfolio|
|US6115697||Feb 19, 1999||Sep 5, 2000||Dynamic Research Group||Computerized system and method for optimizing after-tax proceeds|
|US6154732||Jul 25, 1997||Nov 28, 2000||Guidedchoice.Com||System for providing investment advice and management of pension assets|
|US6161098||Sep 14, 1998||Dec 12, 2000||Folio (Fn), Inc.||Method and apparatus for enabling small investors with a portfolio of securities to manage taxable events within the portfolio|
|US6240399||Jul 2, 1999||May 29, 2001||Glenn Frank||System and method for optimizing investment location|
|US6393409 *||Oct 31, 1997||May 21, 2002||Morgan Stanley Dean Witter & Co.||Computer method and apparatus for optimizing portfolios of multiple participants|
|US6687681 *||May 28, 1999||Feb 3, 2004||Marshall & Ilsley Corporation||Method and apparatus for tax efficient investment management|
|US7016873 *||Sep 5, 2000||Mar 21, 2006||Charles Schwab & Co., Inc.||System and method for tax sensitive portfolio optimization|
|US7047217 *||Feb 18, 2000||May 16, 2006||Dynamic Taxoptimizer Llc||Computerized system and method for optimizing after-tax proceeds involving options|
|US7249080||Oct 25, 2000||Jul 24, 2007||Upstream Technologies Llc||Investment advice systems and methods|
|US20020143682 *||Nov 27, 2001||Oct 3, 2002||Bergmann Michael D.||Method of ascertaining an efficient frontier for tax-sensitive investors|
|1||ITG Financial Engineering Group, "Capital Gains Taxes and Portfolio Optimization," Jul. 18, 2000, Boston, MA.|
|2||Jean L.P. Brunel, "Tax-Aware Equity Investing", Association for Investment Management and Research.|
|3||Jean L.P. Brunel, "Tax-Aware Portfolio Construction," Association for Investment Management and Research.|
|4||Jean L.P. Brunel, "The Upside-Down World of Tax-Aware Investing," Trusts & Estates (Feb. 1997).|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7729972 *||Dec 6, 2006||Jun 1, 2010||The Bank Of New York Mellon Corporation||Methodologies and systems for trade execution and recordkeeping in a fund of hedge funds environment|
|US7853510 *||Apr 3, 2007||Dec 14, 2010||Itg Software Solutions, Inc.||Method and system for multiple portfolio optimization|
|US7870051 *||Jul 1, 1999||Jan 11, 2011||Fmr Llc||Selecting investments for a portfolio|
|US7873567||Dec 29, 2003||Jan 18, 2011||Asset Trust, Inc.||Value and risk management system|
|US7917424||May 13, 2008||Mar 29, 2011||Bgc Partners, Inc.||Systems and methods for providing financial instruments including contrary positions|
|US8001029 *||Mar 26, 2004||Aug 16, 2011||Ubs Ag||Method and computer program for tax sensitive investment portfolio management|
|US8005740||Oct 31, 2007||Aug 23, 2011||Research Affiliates, Llc||Using accounting data based indexing to create a portfolio of financial objects|
|US8117096 *||Dec 8, 2008||Feb 14, 2012||Q-Biz Solutions, LLC||Private equity accounting and reporting system and method|
|US8170935 *||Oct 21, 2008||May 1, 2012||Nyse Amex Llc||Systems and methods for evaluating the integrity of a model portfolio of a financial instrument|
|US8266038 *||Sep 22, 2011||Sep 11, 2012||Sapere Ip, Llc||Methods, systems, and computer program products for providing low risk portable alpha investment instruments|
|US8285620 *||Sep 9, 2009||Oct 9, 2012||Westpeak Global Advisors, LLC||Methods and systems for building and managing portfolios based on ordinal ranks of securities|
|US8332299 *||Dec 19, 2008||Dec 11, 2012||Wells Fargo Bank, N.A.||Investment policy tool|
|US8352347 *||Dec 29, 2009||Jan 8, 2013||Athenainvest, Inc.||Investment classification and tracking system using diamond ratings|
|US8374937 *||Nov 16, 2009||Feb 12, 2013||Research Affiliates, Llc||Non-capitalization weighted indexing system, method and computer program product|
|US8374939 *||Apr 1, 2010||Feb 12, 2013||Research Affiliates, Llc||System, method and computer program product for selecting and weighting a subset of a universe to create an accounting data based index and portfolio of financial objects|
|US8374951 *||Sep 7, 2009||Feb 12, 2013||Research Affiliates, Llc||System, method, and computer program product for managing a virtual portfolio of financial objects|
|US8380604||Jun 19, 2010||Feb 19, 2013||Research Affiliates, Llc||System, method and computer program product for using a non-price accounting data based index to determine financial objects to purchase or to sell|
|US8447687 *||Mar 30, 2010||May 21, 2013||Albert OGRODSKI||Method and system for centralized identity and account controls|
|US8473398 *||Oct 5, 2012||Jun 25, 2013||Westpeak Global Advisors, LLC||Methods and systems for building and managing portfolios based on ordinal ranks of securities|
|US8478788 *||Feb 16, 2010||Jul 2, 2013||Symantec Corporation||Centralized information technology resources analysis system|
|US8589276||Aug 23, 2011||Nov 19, 2013||Research Afiliates, LLC||Using accounting data based indexing to create a portfolio of financial objects|
|US8606618 *||Mar 31, 2011||Dec 10, 2013||Mindjet Llc||Method and system for web-based idea prediction market investment|
|US8635141||Dec 10, 2010||Jan 21, 2014||Itg Software Solutions, Inc.||Method and system for multiple portfolio optimization|
|US8688575 *||Feb 7, 2011||Apr 1, 2014||Neal Robert Steiner||Customizable investment fund and investing education|
|US8694402 *||Aug 23, 2012||Apr 8, 2014||Research Affiliates, Llc||Using accounting data based indexing to create a low volatility portfolio of financial objects|
|US8694406||Mar 23, 2012||Apr 8, 2014||Athenainvest, Inc.||Strategy market barometer|
|US8732058||Oct 17, 2009||May 20, 2014||Cfph, Llc||Methods and systems for combining securities and carbon credits|
|US8732061||Dec 27, 2001||May 20, 2014||Bgc Partners, Inc.||Creating and trading dynamic securities|
|US8756125 *||Jul 31, 2006||Jun 17, 2014||Intuit Inc.||Method and system for providing remote financial assistance|
|US8930255||Sep 15, 2012||Jan 6, 2015||Cfph, Llc||Methods and systems for combining securities and carbon credits|
|US20050071266 *||Dec 29, 2003||Mar 31, 2005||Eder Jeff Scott||Value and risk management system|
|US20050234793 *||Mar 26, 2004||Oct 20, 2005||Renato Staub||Method and computer program for tax sensitive investment portfolio management|
|US20100042367 *||Mar 5, 2008||Feb 18, 2010||Rolls-Royce Plc||Analysis method|
|US20100063942 *||Sep 7, 2009||Mar 11, 2010||Research Affiliates, Llc||System, method, and computer program product for managing a virtual portfolio of financial objects|
|US20100169237 *||Dec 29, 2009||Jul 1, 2010||Athenainvest, Inc.||Investment classification and tracking system using diamond ratings|
|US20100185557 *||Jul 22, 2010||Strategic Capital Network, Llc||Resource allocation techniques|
|US20100191628 *||Apr 1, 2010||Jul 29, 2010||Research Affiliates, Llc||System, Method and Computer Program Product for Selecting and Weighting a Subset of a Universe to Create an Accounting Data Based Index and Portfolio of Financial Objects, as amended|
|US20100250411 *||Sep 30, 2010||Ogrodski Albert||Method and system for centralized identity and account controls|
|US20100287116 *||Nov 16, 2009||Nov 11, 2010||Research Affiliates, Llc||Non-capitalization weighted indexing system, method and computer program product|
|US20110208670 *||Aug 25, 2011||Jpmorgan Chase Bank, N.A.||Execution Optimizer|
|US20110258139 *||Oct 20, 2011||Neal Steiner||Customizable investment fund and investing education|
|US20120016787 *||Jan 19, 2012||Scott Patrick Trease||Methods, Systems, and Computer Program Products for Providing Low Risk Portable Alpha Investment Instruments|
|US20120072372 *||Sep 26, 2011||Mar 22, 2012||Scott Patrick Trease||Methods, Systems and Computer Program Products for Providing Low Risk Portable Alpha Investment Instruments|
|US20120150585 *||Jun 14, 2012||Spigit Inc.||Method and system for web-based idea prediction market investment|
|US20120254068 *||Apr 12, 2012||Oct 4, 2012||Weber Clifford J||Systems and methods for checking model portfolios for actively managed funds|
|US20120310817 *||Aug 16, 2012||Dec 6, 2012||Sapere IP, LLC.||Methods, Systems and Computer Program Products for Providing Low Risk Portable Alpha Investment Instruments|
|US20130124384 *||Dec 31, 2012||May 16, 2013||Optionsxpress Holdings, Inc.||Financial instruments trading system and method|
|US20150012459 *||Oct 26, 2011||Jan 8, 2015||Stanley M. Burnstein||Computerized stock market investment decision-making for individual stocks|
|USRE44098||Aug 8, 2012||Mar 19, 2013||Research Affiliates, Llc||Using accounting data based indexing to create a portfolio of assets|
|USRE44362||Aug 17, 2012||Jul 9, 2013||Research Affiliates, Llc||Using accounting data based indexing to create a portfolio of financial objects|
|WO2013009760A1 *||Jul 10, 2012||Jan 17, 2013||Tapesh Yadav||Systems and methods for investment portfolio management|
|Cooperative Classification||G06Q40/06, G06Q40/10|
|Dec 21, 2001||AS||Assignment|
Owner name: PLACEMARK INVESTMENTS, INC.,TEXAS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:PRUITT, RONALD EARL;REEL/FRAME:012451/0194
Effective date: 20011220
|Jan 9, 2006||AS||Assignment|
Owner name: HORIZON TECHNOLOGY FUNDING COMPANY LLC,CONNECTICUT
Free format text: SECURITY INTEREST;ASSIGNOR:PLACEMARK INVESTMENTS, INC.;REEL/FRAME:017436/0068
Effective date: 20051220
|Jun 14, 2010||AS||Assignment|
Owner name: HORIZON TECHNOLOGY FUNDING COMPANY LLC,CONNECTICUT
Free format text: RELEASE;ASSIGNOR:PLACEMARK INVESTMENTS, INC.;REEL/FRAME:024535/0358
Effective date: 20100428
|Aug 23, 2013||FPAY||Fee payment|
Year of fee payment: 4